Inverted day-night feeding during pregnancy affects the brain health of both maternal and fetal brains through increasing inflammation levels associated with dysbiosis of the gut microbiome in rats

Abstract

Background

In both humans and rodents, inappropriate feeding times during pregnancy can cause maternal metabolic abnormalities, increasing the risk of neurodevelopmental disorders in both the mother and offspring. Using a rat model, this study investigates whether feeding only during the inactive phase in rats leads to anxiety-like behaviors and abnormal brain development in fetuses through gut microbiota imbalance.

Methods

10-week-old female rats in the inactive-phase feeding group (IF group) were first trained for daytime feeding, ensuring that energy intake was statistically insignificant and different from that of the normal diet feeding group (ND group). They were then paired with male rats, and the previous feeding regimen was continued after pregnancy. Anxiety-like behavior was evaluated using the open-field test. Maternal caecal microbiota was analyzed using 16S rRNA sequencing. Enzyme-linked immunosorbent assay (ELISA) measured serum inflammation factors. RT-PCR was employed to assess mRNA levels of integrity genes and inflammatory cytokines in the maternal hippocampi, intestines, fetal brains, and placentae.

Results

There were no statistically significant differences in energy intake or body weight gain between the IF and ND groups. In the open field test, dams in the IF group exhibited anxiety-like behavior, as indicated by fewer entries into and shorter duration in the central zone. Active-phase fasting elevated maternal serum inflammatory cytokine levels and impaired antioxidant capacity. It also increased intestinal permeability and induced gut microbiota dysbiosis, characterized by a decrease in Akkermansia and an increase in Dubosiella. Changes in the expression of intestinal circadian genes and elevated intestinal inflammatory cytokines were observed. Lipopolysaccharide (LPS) translocated into the maternal circulation, activated Toll-like receptor 4 (TLR 4), and passed through the compromised placental barrier into the fetal brain, leading to increased expression of inflammatory cytokines in the fetal brain.

Conclusions

The misalignment between maternal feeding time and the biological clock during pregnancy disrupts the balance of the gut microbiota and peripheral rhythms. The impaired intestinal and placental barriers allow LPS from the gut to infiltrate the maternal hippocampus and fetal brain, increasing inflammation and impacting both maternal and fetal brain health.

Introduction

Time-restricted feeding, a form of intermittent fasting, involves fasting for more than 12 h daily and represents an emerging behavioral nutritional intervention that can assist in treating neuropsychiatric disorders [1]. However, there is ongoing debate regarding whether this 24-h cyclical dietary pattern needs to align with the intrinsic circadian clock, specifically concerning the optimal fasting window [2]. When the eating window aligns with the body's circadian rhythm, TRF can enhance physiological repair processes, balance the gut microbiota, and improve metabolic flexibility [3,4,5]. Conversely, when the two are misaligned, such as inverted day-night feeding, it can lead to a dissociation between peripheral and central circadian clocks, thereby promoting the development of diseases such as obesity, diabetes, and mood disorders [6,7,8]. Thus, meal timing plays a critical role in maintaining normal metabolism. Pregnancy is a very sensitive period, and disturbances in its biological clock can affect the health of both oneself and offspring in both humans and animals [9,10,11]. However, limited research has been conducted on the potential adverse consequences of reversed feeding time during pregnancy.

When feeding rhythms are disrupted, such as through shifts in eating windows or continuous consumption of high-energy foods, these environmental cues give feedback to peripheral clocks, causing clock misalignment and metabolic disturbances [12]. Circadian rhythm disruption is a significant contributor to emotional vulnerability, with detrimental moods further impairing workers' productivity [13]. Moreover, internal rhythm disturbances are strongly associated with anxiety and depression [14]. Such dysregulation can result in the loss of neurons and astrocytes, leading to anxiety- and depression-like behaviors in mice [15].

Changes in feeding windows, including extended eating durations or altered meal timings, have been associated with increased risks of depression and anxiety [16]. Pregnancy is a transformative period for women, characterized by significant physiological and psychological changes [17].

Studies on pregnant women have found that as pregnancy progresses, the continuously enhanced emotional processing ability increases the susceptibility of pregnant women to anxiety disorders [18, 19]. Maternal circadian disruption during pregnancy causes heightened stress and psychological distress, including depression and anxiety [10]. An increase in maternal stress is highly likely to be transmitted to the offspring through the early intrauterine environment. Studies on rodents have found that circadian misalignment related to the pre-pregnancy period and pregnancy is associated with depression and anxiety in male offspring, and these effects persistently influence the physical and behavioral development of the rodents' offspring [11, 20].

In mammalian models, as crucial mediators, the circadian clock and gut microbiota serve as bridges in elucidating how external environmental factors impact the body's metabolism. Both external environmental conditions and internal systemic cycles influence them. In these models, circadian genes interact via interlocked transcription-translation feedback loops and also directly or indirectly engage in metabolic pathways [21]. Studies on rodents have found that the gut microbiota modulates the morphology and function of the brain by producing specific signaling molecules, such as LPS, bile acids, and short-chain fatty acids [22,23,24]. When the gut microbiota is disrupted, the production levels of its metabolites also change, and these changes have been linked to blood–brain barrier integrity and neuroinflammation [25]. In individuals with anxiety and major depressive disorder, gut microbiota dysbiosis has been observed, characterized by a depletion of butyrate-producing anti-inflammatory bacteria and an enrichment of pro-inflammatory bacteria [26]. In addition, the gut microbiota of pregnant women has an impact on both the mother and the fetus. On the one hand, correcting the dysregulated gut microbiota can alleviate maternal pregnancy-related anxiety and depressive emotions [27]. On the other hand, studies on mice have shown that the maternal gut microbiota is crucial for embryonic development [28, 29]. The depletion of the gut microbiota during pregnancy in mice disrupts the formation of thalamocortical axons in embryos [30]. In addition, metabolites originating from the maternal gut of pregnant mice can be transmitted along the maternal gut-placenta-fetus axis in mice, thus leading to intrauterine growth restriction in the fetuses [31].

For pregnant women who work night shifts and those who observe Ramadan fasting, alterations in the eating windows during pregnancy are common phenomena [32, 33]. These women typically consume meals at night, which may increase the risk of delivering small-for-gestational-age infants and have adverse effects on the long-term health of their offspring in adulthood [34, 35]. However, current population studies investigating the impact of such eating habits during pregnancy often involve a reduction in food intake. This makes it unclear whether the observed effects on pregnant women and neonates are entirely attributable to disruptions in feeding schedules [36, 37].

A rat study with a similar research objective to ours found that repeated cycles of Ramadan-like fasting during pregnancy altered placental amino acid transport, leading to low birth weight in the offspring [38]. Given the structural similarities (both being hemochorial placentas) and highly comparable gene expression profiles between rat and human placentas, we selected rats as the experimental model [39]. Additionally, we found no studies investigating whether reversed maternal feeding times during pregnancy disrupt the maternal gut microbiota balance and subsequently affect both the mother and fetal brain development.

Based on the above perspectives, we hypothesize that restricting feeding to the inactive phase in pregnant rats without significantly reducing energy intake may disrupt gut microbiota and circadian rhythm genes, potentially altering the intestinal environment and allowing harmful gut metabolites to affect maternal brain function via the gut-brain axis and fetal brain development via the gut-placenta axis.

Materials and methods

Animal experiment

First experiment: Twenty 6-week-old female Sprague–Dawley rats and ten male rats were purchased from Vital River Laboratory Animal Technology Company (Beijing, China). The rats were housed under a controlled condition with a constant temperature of 22 ± 2 °C, humidity of 50 ± 5%, and a 12-h light–dark cycle (Turn on the light at 9:00 and turn off the light at 21:00) for four weeks, with unrestricted access to water throughout the experiment. The rats were fed a purified maintenance diet (AIN93M) purchased from Beijing Keao Xieli Feed Company.

Before the experiment, the 10-week-old animals were randomly divided into two groups based on body weight: the ND and the IF, with the ND group having unrestricted access to food daily, while the IF group was fed from 9:00 to 21:00 and fasted from 21:00 to 9:00, giving them a 12-h daily feeding period. To ensure equal energy intake between the IF group and the ND group, the IF group underwent a three-week training period. This training duration was determined based on previous research [40], aiming to help the experimental animals adapt to the new feeding-fasting cycle. The daily energy intake for the IF group was calculated and adjusted to have no statistically significant difference from the ND group. Once intake was balanced, the female and male rats were paired in a 2:1 ratio. A vaginal smear was taken from the female rats, and the presence of sperm was recorded as a successful mating, marking gestational day 0.5 (GD0.5). Once pregnancy was confirmed, each female rat was housed individually with separate access to food and water. Two female rats in the IF group were excluded due to difficulties in conception. From GD0.5, body weight was measured every six days and twice daily, at 9:00 and 21:00, and the average of the two measurements was recorded as the rat’s body weight for that day. On GD19.5, female rats subjected to 12 h of fasting were euthanized via intraperitoneal injection of sodium pentobarbital at a dose of 40 mg/kg. Under sterile conditions, maternal organs, including the ileum, colon, cecal contents, hippocampus, hypothalamus, placentae, and fetal brains, were collected. These samples were immediately frozen in liquid nitrogen and stored at −80 °C for long-term preservation. In the subsequent experiments, one female and one male placenta, along with the corresponding fetal brains, were collected from each dam. The selection was based on the median fetal body weight within each litter.

In the second experiment, nine-week-old forty-eight female rats and twenty-four male rats were purchased from the same company. After one week of acclimation feeding ,the experiment was initiated. Dietary training was initiated after adaptive feeding. The procedures during pregnancy were the same as above. Tissue collection was conducted at fixed time points from GD19.5 to 20.5, with samples collected every 6 h. The study of laboratory animals was approved by the Harbin Medical University Institutional Animal Care Committee.

Biochemical analysis

After serum samples were separated from whole blood, total antioxidant capacity (T-AOC) and reduced glutathione (GSH) were measured in serum using standard enzyme colorimetric methods (Nanjing Jiancheng, Nanjing, China). Tissue levels of T-AOC and GSH were measured with the above reagent kits according to the manufacturer's instructions.

Open field test

Rats were placed in the center of a black PVC square arena (100 cm × 100 cm × 50 cm) and allowed to explore the quiet environment for 10 min. Between each experiment, the arena was cleaned with 70% ethanol. The arena floor was divided into 16 squares, with the central four squares defined as the center area and the surrounding 12 squares as the peripheral area. During testing, rats were placed in the central square, and their behavior was tracked for 5 min using the Smart Behavioral Analysis System (Panlab, USA). Data recorded included the number of grid crossings, total distance traveled, and time spent in the center area.

Elevated plus maze

Pregnant rats were placed in the testing room to acclimate to the new environment, and the maze was thoroughly cleaned. The standard EPM apparatus consists of two closed arms (50 × 10 × 30 cm) and two open arms (50 × 10 cm) arranged in a cross shape. The central platform is 10 × 10 cm and elevated 45 cm above the ground. Each test session lasted 5 min. After each test, the apparatus was cleaned with 75% ethanol. Rats' activities were recorded using Smart 3.0 software, including entries into each arm and time spent in each area. Data were analyzed using SPSS to compare differences between groups and reveal potential biological significance.

Histological examination

The placentae were fixed in 4% paraformaldehyde. After fixation, the tissues were dehydrated in an alcohol gradient, treated with n-butanol and xylene, and embedded in paraffin. The samples were sectioned at a thickness of 5 μm using a paraffin microtome. Sections were then stained with hematoxylin and eosin (H&E), mounted, and scanned. ImageJ software was used to analyze the morphological features of the placental tissues.

RNA isolation and quantitative real-time PCR

RNA was isolated from tissue samples using the FreeZol Reagent kit (Vazyme, China), following the manufacturer's instructions. A measured amount of tissue was used for RNA extraction, and 1 μg of total RNA was then reverse-transcribed into cDNA using the HiScript III RT SuperMix for qPCR (+gDNA wiper, Vazyme, China) cDNA kit according to the instructions. Primers were designed based on the CDS regions of the target genes in the NCBI database and verified using BLAST; target gene sequences are provided in the supplementary materials. Quantitative PCR was performed with SYBR Green dye (Mei5bio, China) using a LightCycler® 480 Instrument (Roche). GAPDH served as the reference gene for placental and fetal brain samples, while β-actin was used as the reference for maternal tissues, including the hippocampus, intestine, and hypothalamus. The relative expression levels of target genes were quantified using the 2^−ΔΔCt method.

16S rRNA sequencing

Cecal content samples were collected at GD19.5, and bacterial DNA was isolated using a commercial extraction kit, following the manufacturer's instructions. After genomic DNA extraction, the V3-V4 region of the 16S rDNA was amplified using the forward primer 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and the reverse primer 806R (5′-GGACTACHVGGGTWTCTAAT-3′). The amplified products were verified using agarose gel electrophoresis. Qualified samples were sequenced using the Illumina NextSeq 2000 PE300 platform (provided by Majorbio, Shanghai, China).

The Shannon index was calculated to assess the microbial diversity of the cecal contents. Principal Coordinate Analysis (PCoA) based on the Bray–Curtis distance algorithm was used to evaluate the similarity of microbial communities at the phylum and genus levels, with the Adonis test used to determine if the differences between groups were statistically significant. Linear Discriminant Analysis Effect Size (LEfSe) was applied to identify bacterial taxa with substantial differences in abundance from phylum to genus levels between groups (LDA > 2, p < 0.05). Functional predictions of the microbiome in the samples were made using Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt2) analysis. All analyses were conducted on the Majorbio Cloud Platform (https://cloud.majorbio.com).

Enzyme-linked immunosorbent assay (ELISA)

A 50 mg tissue sample was homogenized in phosphate-buffered saline PBS at a 1:9 volume ratio. The homogenate was centrifuged at 3000 rpm for 15 min at 4 °C, and the supernatant was collected for analysis. Serum samples were thawed directly for testing. Following the manufacturers’ instructions, ELISA tests were conducted for various biomarkers. The ELISA kits for interleukin-6 (IL-6), tumor necrosis factor-alpha (TNF-α), and interleukin-1 beta (IL-1β) were obtained from Solarbio Science & Technology (Beijing, China), and the LPS ELISA kits were provided by Jianglai Biological (Shanghai, China).

Statistical analysis

All data in this study were expressed as mean ± SEM. Data analysis was performed using SPSS 25.0 software. Initially, normality and homogeneity of variance of the data were tested. For normally distributed data, Student’s t-test was applied; for non-parametric data, the Mann–Whitney U test was used. Chi-square tests were applied to percentage data, and non-parametric tests were used for microbiome data. Energy intake data were compared using the generalized estimating equation. The strength of correlations was determined using Spearman's correlation coefficients. GraphPad Prism 9.0 and RStudio were used to present the results, with statistical significance set at p < 0.05 (ns indicates p > 0.05, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Results

Fasting during the active phase did not alter maternal weight during pregnancy but reduced the weight of the fetuses, placentae, and fetal brains

Since rats are nocturnal animals, their daytime food intake constitutes only about 20% of the total daily intake [41]. Therefore, prior to pregnancy, nulliparous rats were trained to consume their entire daily food intake during the inactive phase, as shown in Fig. 1a. After this pre-pregnancy training, energy intake during pregnancy did not differ between the IF group and the ND group (Fig. 1b). Likewise, changes in maternal weight during pregnancy and total weight gain throughout pregnancy were not statistically significantly different (p > 0.05) (Fig. 1c, d). When examining the wet weight of maternal tissues, the hippocampi in the IF group were lighter compared to the ND group, whereas the weight of the hearts, kidneys, spleens, and livers showed statistically insignificant changes (Fig. 1e). Next, we assessed fetal developmental conditions. The number of fetuses (Fig. 1f) and sex ratios (Fig. 1g) in the IF group showed statistically insignificant differences compared to the ND group. However, the weight of both male and female fetuses was statistically significantly reduced in the IF group (Fig. 1h). Placental weight, an important predictor of fetal development [42, 43], was statistically significantly reduced in female fetuses from the IF group, while males showed a slight but not statistically significant reduction in placental weight (Fig. 1i, j). Finally, the weight of fetal brains was examined, and only females in the IF group exhibited reduced whole-brain weight compared to the ND group (Fig. 1k).

Fig. 1

Effects of active-phase fasting on maternal and fetal weight during pregnancy. a Experimental design and timeline. Formal experimentation began after confirming pregnancy through vaginal smears, marking GD0.5. Maternal energy intake was recorded every 24 h until GD18.5 (ND group, n = 10; IF group, n = 8). b Feeding data was collected over 17 complete 24-h cycles. c Maternal body weight at GD0.5, GD6.5, GD12.5, and GD18.5. d Maternal weight gain during pregnancy. e Average wet weight of maternal organs at GD19.5, including hippocampi, heart, kidneys, spleen, and liver from top to bottom. f–k Developmental status of the fetus in ND and IF groups (ND group, n = 141 fetuses; IF group, n = 92 fetuses): f Average fetus number per litter. g Fetal sex ratio. h Weight of the fetuses. i Average wet weight of the placenta. j Placental efficiency (calculated as fetal weight/placenta weight). k Total fetal brain weight. Results for panels (b–f and h–k) were presented as mean ± SEM. Statistical significance was denoted as follows: ns indicated no statistical difference (p > 0.05); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Fasting during the active phase increased anxiety-like behavior in pregnant rats, which was associated with elevated inflammation levels in serum and the hippocampus

Adverse dietary habits during pregnancy can heighten the risk of emotional disorders in rats [44]. The open field test, commonly used to assess anxiety-like behavior in experimental animals [45], revealed anxiety-like behavior in the IF group in GD18.5, characterized by reduced time spent in the central zone (48.70 ± 9.29 vs. 12.80 ± 2.00 s, p = 0.008) (Fig. 2a), increased time in the peripheral zone (251.30 ± 9.28 vs. 287.20 ± 2.00 s, p = 0.008) (Fig. 2b), and fewer entries into the central zone (Fig. 2c), although total distance traveled showed statistically insignificant difference (Fig. 2d). Additionally, we conducted an elevated plus maze (EPM) test and found that, compared to the ND group, IF dams showed a reduced number of entries into the open arms and spent less time in the open arms (Supplementary Fig. 1a–d). These results further support that active-phase time-restricted feeding induces anxiety-like behavior in pregnant rats. Serum inflammatory levels and antioxidant capacity are critical indicators of anxiety [46, 47]. The IF group displayed statistically significantly higher serum levels of TNF-α (205.40 ± 17.14 vs. 319.04 ± 31.35 pg/ml, p = 0.006), IL-6 (124.42 ± 16.25 vs. 195.60 ± 27.61 pg/ml, p = 0.039), and IL-1β (116.66 ± 2.73 vs. 126.10 ± 3.15 pg/ml, p = 0.041), along with lower levels of T-AOC (931.06 ± 75.53 vs. 400.23 ± 48.14 μM, p < 0.0001) and GSH (43.22 ± 4.59 vs.14.29 ± 4.57 μM, p = 0.001) (Fig. 2e–i). The hippocampus, a key region in regulating emotions [48], showed increased mRNA expression of inflammatory cytokines, including TNF-α (2.01-fold, p = 0.032), IL-6 (2.01-fold, p = 0.015), IL-1β (1.72-fold, p = 0.037), nucleotide-binding, oligomerization domain-like receptor family pyrin domain containing 3 (NLRP3) (1.84-fold, p = 0.018), and inducible nitric oxide synthase (iNOS) (1.90-fold, p = 0.009), in the IF group (Fig. 2j). Furthermore, T-AOC and GSH (Fig. 2k, l) levels in the hippocampus were reduced in the IF group (30.62 ± 1.12 μmol/g protein vs. 21.64 ± 2.75 μmol/g protein, p = 0.014; 49.98 ± 4.90 μmol/g protein vs. 31.84 ± 4.03 μmol/g protein, p = 0.013, respectively) compared with the ND group. Spearman correlation analysis revealed that serum and hippocampal inflammation levels were positively correlated with peripheral zone spending time but negatively correlated with the number of entries into and time spent in the central zone. Conversely, serum and hippocampal T-AOC levels were negatively correlated with spending time in the peripheral zone and positively correlated with central zone entries and time spent. Although serum GSH levels were strongly correlated with anxiety indicators, hippocampal GSH levels showed a statistically insignificant correlation with anxiety indices (Fig. 2m). In summary, the elevated levels of inflammation markers, including IL-6, TNF-α, IL-1β, and iNOS, in the serum and hippocampus of the IF group strongly correlated with anxiety-like behavior indicators (|r|> 0.6). Therefore, increased inflammation in pregnant rats may be one of the contributing factors to their anxiety-like behaviors.

Fig. 2

Active-phase fasting-induced anxiety-like behavior in pregnant rats, associated with increased serum and hippocampal inflammation. a–d Anxiety-like behavior indicators in pregnant rats on gestational day 18.5 (ND group, n = 7; IF group, n = 6): a Time spent in the central area (seconds). b Time spent in the peripheral area (seconds). c Number of entries into the central region. d Total distance traveled (meters). e–i Serum pro-inflammatory cytokines and antioxidant levels on GD19.5 (ND group, n = 8; IF group, n = 7): e TNF-α levels. f IL-6 levels. g IL-1β levels. h T-AOC. i GSH levels. j–l Hippocampal pro-inflammatory cytokine mRNA levels and antioxidant capacity on GD19.5: j Relative mRNA expression levels of hippocampal pro-inflammatory cytokines (TNF-α, IL-6, IL-1β, NLRP3, iNOS) (ND group, n = 8; IF group, n = 7). k Hippocampal T-AOC level. l Hippocampal GSH levels (n = 8 each group). m Correlation heatmap illustrating the relationship between serum/hippocampal inflammation, antioxidant levels, and anxiety-like behavior indicators. Yellow indicated positive correlations, and purple indicated negative correlations. Results in panels a–l were presented as mean ± SEM. Statistical significance was denoted as follows: ns indicated no statistical difference (p > 0.05); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Impaired intestinal barrier, elevated inflammation, and increased LPS levels linked to Altered cecal microbiota in fasting-phase maternal rats

As a critical organ for food digestion and absorption, the intestine plays a vital role in maintaining systemic health. Increased intestinal permeability can lead to bacterial translocation and the entry of gut-derived antigens into the bloodstream [49], which may contribute to the onset and progression of emotional disorders. To investigate this, we analyzed the expression levels of tight junction proteins, including zonula occludens-1 (ZO-1), occludin (Ocln), claudin 1 (Cldn1), and mucin 2 (Muc2), in the ileum and colon. The results showed that the relative mRNA expression levels of tight junction proteins were statistically significantly reduced in the ileum and colon of the IF group, and the mRNA level of Muc2 was also markedly decreased (Fig. 3a, c). In the ileum, ZO-1 was reduced to 0.46-fold (p = 0.001), Ocln to 0.19-fold (p = 0.005), Cldn1 to 0.29-fold (p = 0.003), and Muc2 to 0.45-fold (p = 0.008). In the colon, ZO-1 decreased to 0.42-fold (p = 0.005), Ocln to 0.28-fold (p = 0.009), Cldn1 to 0.38-fold (p = 0.012), and Muc2 to 0.42-fold (p = 0.003).

Fig. 3

Active-phase Fasting reduced maternal intestinal integrity, increased intestinal inflammation, and altered gut microbiota characteristics. a Relative mRNA expression levels of ileal tight junction proteins (ZO-1, Ocln, Cldn1) and mucin (Muc2). b Relative mRNA expression levels of ileal inflammatory cytokines (TLR 4, TNF-α, IL-6, IL-1β, NLRP3). c Relative mRNA expression levels of colonic tight junction proteins (ZO-1, Ocln, Cldn1) and mucin (Muc2). d Relative mRNA expression levels of colonic inflammatory cytokines (TLR 4, TNF-α, IL-6, IL-1β, NLRP3) (ND group: n=8; IF group: n=7). e Comparison of bacterial abundance and sequence counts in cecal contents. f Shannon index-based bacterial α-diversity at the phylum and genus levels. g PCoA at the phylum and genus levels. h Dysbiosis index at the phylum and genus levels. ND group: n = 7; IF group: n = 6. Data in a–f and h were expressed as mean ± SEM. Statistical significance was denoted as follows: ns indicated no statistical difference (p > 0.05); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Additionally, compared to the ND group, the IF group showed statistically significantly elevated mRNA expression levels of pro-inflammatory cytokines, including TLR 4 (2.74-fold, p = 0.009), TNF-α (2.18-fold, p = 0.015), IL-6 (2.72-fold, p < 0.0001), IL-1β (2.62-fold, p = 0.001), and NLRP3 (2.58-fold, p = 0.038) in the ileum and TLR 4 (2.54-fold, p = 0.001), TNF-α (2.20-fold, p = 0.011), IL-6 (2.88-fold, p = 0.027), IL-1β (2.24-fold, p = 0.008), and NLRP3 (2.15-fold, p < 0.0001) in the colon (Fig. 3b, d). These findings suggested that fasting during the active phase increased intestinal permeability, weakened barrier function, and induced intestinal inflammation in maternal rats.

To explore whether these changes were associated with intestinal microbiota, we analyzed the characteristics of the cecal microbiota. There were statistically insignificant differences between the ND and IF groups in the total number of bases and gene sequences (Fig. 3e). At both the phylum and genus levels, microbial diversity tended to decrease in the IF group, but the changes were not statistically significant (Fig. 3f). However, PCoA revealed distinct clustering between the two groups, supported by Adonis analysis (Fig. 3g). Furthermore, the microbiota dysbiosis index was statistically significantly higher in the IF group compared to the ND group (Fig. 3h). At the phylum level, the proportion of Proteobacteria was increased, while Verrucomicrobiota was decreased in the IF group, both with statistical significance (p < 0.05, Fig. 4a). At the genus level, 19 genera were altered in the IF group: 10 genera (Akkermansia, Ruminococcus_torques_group, Romboutsia, unclassified_f__Lachnospiraceae, Ruminococcus_gauvreauii_group, Faecalitalea, Turicibacter, unclassified_f__Peptostreptococcaceae, Christensenella, Candidatus_Stoquefichus) were downregulated, while 9 genera (Dubosiella, Bilophila, norank_f__Oscillospiraceae, NK4A214_group, norank_f__Eubacterium_coprostanoligenes_group, norank_f__Muribaculaceae, Lachnospiraceae_UCG-010, unclassified_f__Desulfovibrionaceae, unclassified_f__Christensenellaceae) were upregulated (Fig. 4b). A phylogenetic cladogram visualized taxonomic changes across phyla, classes, orders, families, genera, and species, highlighting taxa with LDA scores > 2 using LEfSe analysis (Fig. 4c). Functional prediction of microbiota using PICRUST2 revealed statistically significant changes in 24 pathways in the IF group: 15 pathways were upregulated, while 9 were downregulated. Notably, the LPS biosynthesis pathway was the most prominent and garnered particular attention (Fig. 4d). To further clarify the relationship between microbiota and inflammation, we analyzed the correlation between 19 differentially abundant genera and inflammatory cytokines or antioxidant capacity in the serum and hippocampus. Strong negative correlations (r < −0.6) were observed between hippocampal inflammation levels and genera such as Akkermansia, Candidatus_Stoquefichus, Faecalitalea, Romboutsia, Ruminococcus_gauvreauii_group, Ruminococcus_torques_group, and unclassified_f__Lachnospiraceae. Conversely, positive correlations (r > 0.6) were observed with unclassified_f__Christensenellaceae, norank_f__Oscillospiraceae, NK4A214_group, Lachnospiraceae_UCG-010, and Dubosiella (Fig. 4e). We also performed a Spearman correlation analysis between intestinal barrier indices, inflammation levels, and microbiota (Fig. 4f). Functional predictions suggested an upregulation of the glutathione metabolism pathway. However, a statistically insignificant correlation was found between hippocampal GSH levels and anxiety indices. Thus, we thought that microbiota primarily contributed to LPS-mediated inflammatory responses rather than GSH antioxidant processes in regulating anxiety-like behaviors.

Fig. 4

Active-phase Fasting Altered Cecal Microbiota Composition in Pregnant Rats and Increased the LPS Biosynthesis Pathway. a Taxonomic composition of the cecal microbiota at the phylum level. b Heatmap of relative expression levels of differentially abundant genera, with data normalized for comparison. c Microbial taxa unique to the active-phase fasting group were identified using LEfSe analysis. d Functional predictions of bacterial communities based on PICRUSt2. e Circular Spearman correlation heatmap showing the relationships between differentially abundant species and levels of serum/hippocampal inflammatory cytokines and antioxidant capacities. f Correlation between differentially abundant genera and markers of ileal/colonic integrity and inflammatory factors, depicted using Spearman correlation. In heatmaps e–f, yellow indicated positive correlations, and purple indicated negative correlations. Results in panel d were presented as mean ± SEM. Statistical significance was denoted as follows: ns indicated no statistical difference (p > 0.05); *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

To determine whether active-phase fasting influenced intestinal LPS levels, we measured LPS content in cecal contents. Compared to the ND group, the IF group exhibited statistically significantly higher LPS levels (151.46 ± 15.91 vs. 239.61 ± 6.45 ng/g, p = 0.0003) (Fig. 5a). Increased intestinal permeability in the IF group allowed more LPS to translocate into the serum (1.08 ± 0.12 vs. 2.43 ± 0.37 ng/ml, p = 0.010) (Fig. 5b) and subsequently into the hippocampus via circulation (0.10 ± 0.01 ng/mg protein vs. 0.18 ± 0.01 ng/mg protein, p < 0.0001) (Fig. 5c). Correlation analysis showed that nine genera negatively regulated LPS levels, including Akkermansia, Ruminococcus_torques_group, Romboutsia, unclassified_f__Lachnospiraceae, Ruminococcus_gauvreauii_group, Turicibacter, unclassified_f__Peptostreptococcaceae, Christensenella, and Candidatus_Stoquefichus, conversely, five genera positively regulated LPS levels, including Dubosiella, Bilophila, norank_f__Oscillospiraceae, NK4A214_group, and norank_f__Muribaculaceae (Fig. 5d–q).

Fig. 5

Active-phase fasting increased the level of LPS, which was associated with gut microbiota. a LPS levels in the cecum of pregnant rats. b LPS levels in serum. c LPS levels in the hippocampus. d–r Correlation between differential cecal genera and LPS levels, analyzed by Spearman correlation: d Akkermansia, e Ruminococcus torques group, f Romboutsia, g unclassified_f__Lachnospiraceae, h Dubosiella, i Bilophila, j Ruminococcus gauvreauii group, k norank_f__Oscillospiraceae, l NK4A214 group, m norank_f__Muribaculaceae, n Turicibacter, o unclassified_f__Peptostreptococcaceae, p Christensenella, q Candidatus Stoquefichus, r Relative expression of hippocampal TLR 4 mRNA. The results in panels a–c and r were presented as mean ± SEM, with ND group n = 8 and IF group n = 7. Panels d–r represented ND group = 7 and IF group n = 6. *p < 0.05, **p < 0.01, ***p < 0.001

As a key receptor for LPS, TLR 4 induces the expression of inflammatory cytokines upon activation [50]. We observed increased hippocampal TLR 4 (2.34-fold, p = 0.006) expression in the IF group (Fig. 5r), further demonstrating that fasting during the active phase elevated maternal LPS levels and activated systemic inflammatory responses.

Maternal active-phase fasting altered intestinal circadian gene expression and was associated with barrier dysfunction and inflammation

Peripheral circadian clocks regulate physiological activities, such as nutrient metabolism, under the influence of feeding schedules and central clock control [51]. Disruption of intestinal circadian genes can lead to gut homeostasis imbalance, potentially contributing to disease development [52]. To examine this, we evaluated the expression of circadian genes in the ileum and colon. In the ileum, active-period fasting resulted in a statistically significant downregulation of brain and muscle aryl hydrocarbon receptor nuclear translocator-like protein 1 (Bmal1) (0.48-fold, p = 0.007), period circadian regulator 2 (per2) (0.47-fold, p = 0.018), cryptochrome 1 (cry1) (0.42-fold, p = 0.019), cryptochrome 2 (cry2) (0.39-fold, p = 0.006), and nuclear receptor subfamily 1, group D member 1 (Rev-Erbα) (0.32-fold, p = 0.008) mRNA levels, while retinoid receptor-related orphan receptor alpha (Rorα) (2.68-fold, p = 0.003) and nicotinamide phosphoribosyltransferase (nampt) (4.20-fold, p = 0.001) expression levels were notably increased. Statistically insignificant changes were observed in the relative expression of circadian locomotor output cycles kaput (clock), period circadian regulator 1 (per1), or albumin D-site-binding protein (DBP) (Fig. 6a). Similarly, in the colon, the expression of clock (0.43-fold, p = 0.031), Bmal1 (0.64-fold, p = 0.03), per2 (0.27-fold, p = 0.011), cry1 (0.69-fold, p = 0.03), and Rev-Erbα (0.42-fold, p = 0.002) was statistically significantly reduced, whereas Rorα (1.94-fold, p = 0.035) and nampt (3.61-fold, p = 0.02) levels were elevated (Fig. 6b). Since central circadian rhythms regulate peripheral circadian genes [53], we analyzed circadian gene expression in the hypothalamus to determine whether active-period fasting affects intestinal circadian genes via central clock modulation. Statistically insignificant alterations in hypothalamic circadian gene expression were detected (Supplementary Fig. 2a–j). Given the importance of intestinal circadian rhythms in maintaining gut stability [51], we investigated the correlation between circadian gene expression, intestinal barrier function, and inflammation. This analysis provided indirect evidence linking alterations in circadian genes with intestinal barrier dysfunction and inflammatory responses (Fig. 6c).

Fig. 6

Active-phase fasting altered the levels of rhythmic genes in the ileum and colon of dams. a Relative mRNA expression levels of rhythmic genes in the ileum, including clock, Bmal1, per1, per2, cry1, cry2, Rev-Erbα, Rorα, nampt, and DBP (ND group: n = 8; IF group: n = 7). b Relative mRNA expression levels of rhythmic genes in the colon, including clock, Bmal1, per1, per2, cry1, cry2, Rev-Erbα, Rorα, nampt, and DBP (ND group: n = 8; IF group: n = 7). The results in panels a and b were presented as mean ± SEM. c Heatmap showing correlations between rhythmic genes and intestinal inflammatory factors in the ileum and colon, where yellow indicated a positive correlation, and purple indicated a negative correlation. "ns" indicated no statistical significance (p > 0.05), and superscript asterisks denoted statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001

To further illustrate the effects of fasting during the active phase on maternal intestinal circadian genes, we examined the 24-h oscillation patterns of circadian genes in the ileum, colon (Fig. 7), and SCN (Supplementary Fig. 3) of IF and ND dams at Zeitgeber time (ZT) 0, 6, 12, and 18 h. Consistent with our previous findings, the SCN of IF dams did not exhibit statistically significant changes in response to the feeding-fasting cycle (Supplementary Fig. 3a–j). However, the circadian gene expression in the intestine was influenced by the feeding cycle and became desynchronized from the central clock, as evidenced by a loss of normal rhythmic oscillations. Moreover, compared with the ND group, the IF group showed reduced expression levels of per2, cry1, and Rev-Erbα and increased expression levels of nampt both in the ileum and colon. Notably, the circadian rhythmicity of the cry1 gene was markedly disrupted in the IF group (Fig. 7a, b).

Fig. 7

Active-phase feeding disrupted clock gene rhythmicity in ileum and colon. a Relative mRNA expression levels of rhythmic genes in the ileum at ZT 0, 6, 12, 18 h, including clock, Bmal1, per1, per2, cry1, cry2, Rev-Erbα, Rorα, nampt, and DBP. b Relative mRNA expression levels of rhythmic genes in the colon at ZT 0, 6, 12, 18 h, including clock, Bmal1, per1, per2, cry1, cry2, Rev-Erbα, Rorα, nampt, and DBP (n = 6 each group). Data were represented as mean ± SEM for each group

In summary, fasting during the active phase disrupted the expression of circadian genes in the maternal intestine but did not statistically significantly affect the central circadian clock.

Maternal active-phase fasting during pregnancy led to altered placental morphology, increased inflammation, and reduced antioxidant capacity

The placenta serves as a vital connection between the mother and fetus, and maternal stress can disrupt placental structure, adversely affecting early embryonic development. H&E staining revealed that the labyrinth zone proportion in the placenta was statistically significantly reduced in the IF group, indicating incomplete placental development (Fig. 8a, b). Additionally, higher levels of LPS were detected in the placenta of the IF group compared with the ND group (Female: 0.04 ± 0.003 vs. 0.07 ± 0.01 ng/mg protein, p = 0.0005; Male: 0.04 ± 0.005 vs. 0.07 ± 0.01 ng/mg protein, p = 0.028) (Fig. 8c), accompanied by decreased mRNA expression of tight junction proteins ZO-1 (Female: 0.58-fold, p = 0.0097; Male: 0.62-fold, p = 0.006), Cldn1 (Female: 0.58-fold, p = 0.044; Male: 0.63-fold, p = 0.0496), and Ocln (Female: 0.65-fold, p = 0.004; Male: 0.60-fold, p = 0.018), which resulted in increased placental permeability (Fig. 8d). Elevated levels of TLR 4 (2.87-fold, p = 0.003) in the placenta in the IF group, suggesting a response to maternal LPS exposure, were associated with increased mRNA expression of pro-inflammatory factors, including TNF-α (2.67-fold, p = 0.029), IL-6 (2.52-fold, p = 0.011), IL-1β (1.86-fold, p = 0.010), NLRP3 (1.76-fold, p = 0.025), and iNOS (1.57-fold, p = 0.011). Notably, these inflammatory changes were more pronounced in female fetuses compared to males (Fig. 8e, g). Moreover, the total antioxidant capacity of the placenta belonging to females and males was reduced (Female: 161.97 ± 4.96 vs. 131.92 ± 7.37 μmol/g protein, p = 0.005; Male: 159.7 ± 5.13 vs. 131.23 ± 10.38 μmol/g protein, p = 0.026), and GSH levels decreased (Female: 59.71 ± 7.84 vs. 16.53 ± 2.59 μmol/g protein, p = 0.005; Male: 42.07 ± 5.86 vs. 23.71 ± 3.52 μmol/g protein, p = 0.010) (Fig. 8f, h). In summary, maternal active-phase fasting during pregnancy resulted in placental developmental defects, increased permeability, elevated inflammation, and diminished antioxidant capacity. These changes were more pronounced in female fetuses than in males.

Fig. 8

Fasting during the active phase increased placental LPS, inflammation, altered morphology, and reduced antioxidants. a Representative images of H&E-stained placental sections for tissue morphology analysis (5 pregnant rats per group, including 5 female and 5 male fetuses). The area below the blue line represented the labyrinth zone, while the region between the red and blue lines represented the basal zone. b Analysis of placental junction zone, labyrinth area, and the labyrinth/junction zone ratio. c Placental LPS levels. d Placental integrity gene mRNA levels: left for females, right for males—genes from left to right: ZO-1, Cldn1, Ocln. e–h Placental inflammatory factors (TLR 4, TNF-α, IL-6, IL-1β, NLRP3, iNOS) mRNA levels and antioxidant capacity (T-AOC and GSH). (ND group: n = 7, IF group: n = 6) e Placental inflammatory factor mRNA levels in female fetuses. f Placental antioxidant levels in female fetuses. (g) Placental inflammatory factors in male fetuses. h Placental antioxidant levels in male fetuses (T-AOC, GSH). Results in panels b–h were presented as mean ± SEM. "ns" indicated p > 0.05 (no statistical difference), and asterisks (*) denoted statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Maternal active-phase fasting increased inflammation and reduced antioxidant capacity in the fetal brains, associated with maternal microbiota alterations

Pregnancy is a critical period for embryonic brain development, during which exposure to maternal stressors may disrupt normal development, increasing the risk of neuropsychiatric disorders in adulthood [54]. Our findings indicated that maternal active-phase fasting impaired the placental barrier, potentially exposing fetuses to harmful maternal metabolites. To investigate this, we assessed inflammation levels in the fetal brains to determine whether it was affected by elevated maternal LPS levels. Similar to the placenta results, fetal brain LPS levels were increased in the IF group (Female: 0.11 ± 0.01 vs. 0.23 ± 0.03 ng/mg protein, p = 0.007, Male: 0.11 ± 0.01 vs. 0.18 ± 0.03 ng/mg protein, p = 0.010), regardless of sex (Fig. 9a). The antioxidant capacity of fetal brains was also altered in the IF group (Female: 175.79 ± 13.83 vs. 131.2 ± 9.44 μmol/g protein, p = 0.026, Male: 187.9 ± 11.65  vs. 149.04 ± 7.45 μmol/g protein, p = 0.021), and GSH was reduced in the IF group (Female: 92.58 ± 13.77 vs. 46.09 ± 10.08 μmol/g protein, p = 0.023, Male: 108.41 ± 9.96 vs. 60.61 ± 4.6 μmol/g protein, p = 0.002) (Fig. 9b, c). The mRNA levels of inflammatory markers TNF-α, IL-6, IL-1β, NLRP3, and iNOS in the IF group were statistically significantly higher compared to the ND group, with the changes in inflammation levels in female fetal brains being more pronounced than in males, including In the IF group, female brains showed statistically significantly elevated inflammatory markers compared to ND: TLR 4 (9.50-fold, p = 0.0003), TNF-α (3.69-fold, p = 0.0003), IL-6 (10.63-fold, p < 0.0001), IL-1β (2.90-fold, p = 0.0002), NLRP3 (3.67-fold, p < 0.0001), and iNOS (2.08-fold, p = 0.037). In males, the levels were also increased: TLR 4 (4.51-fold, p = 0.005), TNF-α (2.36-fold, p = 0.032), IL-6 (3.75-fold, p = 0.001), IL-1β (3.16-fold, p = 0.0004), NLRP3 (2.62-fold, p = 0.002), and iNOS (1.55-fold, p = 0.003) (Fig. 9d, e). A correlation heatmap further elucidated the relationship between fetal brain inflammation, antioxidant capacity, and maternal gut microbiota. Negative correlations were observed between fetal brain inflammation and genera such as Akkermansia, Romboutsia, unclassified_f__Lachnospiraceae, Ruminococcus_gauvreauli_group, Faecalitalea, Turicibacter, unclassified_f__Peptostreptococcaceae, and Christensenella. Conversely, positive correlations were noted with genera including Dubosiella, Bilophila, NK4A214_group, norank_f__Muribaculaceae, and unclassified_f__Desulfovibrionaceae. Sex-specific differences were observed in the antioxidant capacity of fetal brains: Female brains showed positive correlations with 10 genera and negative correlations with 9 genera. Male brains exhibited positive correlations with 9 genera and negative correlations with 8 genera (Fig. 9f).

Fig. 9

Active-phase fasting increased inflammation levels in the fetal brains, associated with the maternal microbiota. a Fetal brain LPS levels. b, c Fetal brain antioxidant levels (including T-AOC and GSH). d, e Relative mRNA levels of inflammatory factors in the fetal brain, from left to right: TLR 4, TNF-α, IL-6, IL-1β, NLRP3, iNOS. f Correlation heatmap of fetal brain antioxidant capacity and inflammation levels with maternal gut microbiota, with the left side showing female fetal brain data and the right side showing male fetal brain data. Yellow indicated a positive correlation, and purple indicated a negative correlation. For panels a–e, the ND group (n = 7) and IF group (n = 6) were shown. Results were presented as mean ± SEM. "ns" indicated p > 0.05 (no statistical difference), and asterisks (*) indicated statistical significance: *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

Discussion

Feeding time is known to directly influence the composition and abundance of gut microbiota and metabolites [55, 56]. Additionally, it is a key factor in regulating circadian gene expression in intestinal epithelial cells [57]. Disruption of the gut environment can lead to harmful substances accessing the brain via the gut-brain axis, contributing to neuropsychiatric disorders [58]. Furthermore, maternal gut-derived harmful metabolites can affect the intrauterine environment, leading to abnormal early fetal development [31]. This study hypothesizes that active phase fasting during pregnancy disrupts the gut microbiota and intestinal circadian rhythm genes in pregnant rats, increasing intestinal permeability and pro-inflammatory factors. This allows microbiota-derived LPS to enter circulation and activate TLR 4. Then, LPS elevates inflammation in the maternal brain, causing anxiety-like behavior, and crosses the damaged placental barrier into the fetal brain, raising inflammation levels there.

To investigate whether reversed feeding times during pregnancy affect maternal and fetal development, we intentionally restricted feeding to daytime—the inactive phase for rats—throughout the entire gestation period. Previous studies on such dietary habits have mainly focused on 8-h daytime feeding windows, which often result in reduced energy intake due to the short feeding duration [59], which makes it unclear whether the observed effects of such feeding patterns are caused by the altered feeding window or reduced caloric intake. Considering that rats are naturally nocturnal feeders, we trained adult female rats for three weeks before pregnancy to adopt a reversed feeding schedule. Following training, statistically insignificant differences in food intake were observed during the gestation period, thereby excluding the influence of variations in energy intake.

To minimize stress caused by handling, we weighed the pregnant rats only on GD6.5,12.5,18.5 (265.53 ± 21.35 vs. 271.32 ± 15.16 g, 286.71 ± 16.31 vs. 293.15 ± 17.43 g, 302.50 ± 14.81 vs. 318.05 ± 17.76 g, p > 0.05, respectively), statistically insignificant differences in maternal body weight gain were detected among two groups (75.23 ± 4.59 vs. 66.65 ± 4.95 g, p = 0.224). Among non-obese dams, daytime feeding during pregnancy had minimal impact on overall maternal body weight but still resulted in reduced placental, fetuses, and fetal brain weights. This finding aligns with observations in late-gestation Wistar rats subjected to the same feeding pattern [60]. Even though this feeding schedule differs from the previously studied feeding-fasting cycles, it still significantly affects embryonic development [61]. Thus, we believe that reduced fetal weight is associated with inappropriate feeding times during pregnancy, regardless of maternal energy intake restrictions. This effect may result from impaired placental function, which influences embryonic development [61].

Additionally, we observed that maternal rats subjected to active-phase fasting during pregnancy exhibited anxiety-like behaviors on GD18.5. The ventral hippocampus, a critical region for emotion regulation, particularly in anxiety disorders, is highly sensitive to stress stimuli, which can disrupt the balance between excitation and inhibition by altering dendritic and interneuron structures and numbers [62]. The hippocampus is highly susceptible to neuroinflammation and oxidative stress [63, 64]. Pro-inflammatory cytokines can compromise the blood–brain barrier, allowing peripheral cytokines to circulate within the brain [65]. They can also activate astrocytes and microglia to release additional pro-inflammatory cytokines, exacerbating local immune responses [66]. Given the anxiety-like behaviors observed in the maternal rats, we hypothesized that hippocampal inflammation may have occurred. Elevated serum levels of pro-inflammatory cytokines in the active-phase fasting group further indicate a disturbance in gut microbial composition [67]. Thus, we propose that active-phase fasting disrupted the maternal gut environment, increasing vulnerability to pathogenic infections and activating systemic immune responses. Interestingly, a study in rats found that the feeding–fasting cycle may influence the sleep–wake pattern [68]. Sleep deprivation has been shown to induce anxiety in both humans and rodents [69, 70], possibly mediated by neurotransmitter and neuromodulatory systems [71]. However, a study in mice demonstrated that scheduled feeding during the light phase led to a reorganization of the sleep cycle, characterized by reduced sleep during the light period but increased sleep during the dark period, thereby maintaining overall sleep homeostasis [72]. Since our experiment did not include measurements of the sleep–wake cycle in pregnant rats, we are unable to determine whether the inverted feeding schedule disrupted sleep patterns or whether anxiety-like behaviors were mediated through alterations in sleep.

The gut microbiota is influenced by the composition and timing of the host's diet [4]. Studies have shown that feeding mice exclusively during their inactive phase alters the gut microbiota, statistically significantly reducing the abundance of Verrucomicrobia, including Akkermansia [73, 74]. Recently, Akkermansia has been recognized as a probiotic with potential therapeutic applications. It has shown promising effects in improving host immune function, restoring gut microbiota, and alleviating symptoms of neuropsychiatric disorders [75]. In rodent models of anxiety induced by diet or stress, the relative abundance of Akkermansia is consistently reduced [76, 77]. Further research indicates that Akkermansia alleviates disease progression by increasing serum levels of acetate and butyrate, inhibiting microglial activation in the hippocampus, and preventing synaptic phagocytosis, which protects against cognitive decline in sleep-deprived mice [78]. Exogenous supplementation of Akkermansia statistically significantly lowers serum LPS levels and increases gut mucin production [79]. Moreover, Akkermansia-specific outer membrane protein Amuc_1100 interacts with Toll-like receptor 2 in intestinal epithelial cells, modulating serotonin levels in the gut to reduce inflammation and excessive activation of the hypothalamic–pituitary–adrenal axis in anxious mice [80]. Similarly, the Ruminococcus_torques_group is associated with enhancing the intestinal mucosal barrier and reducing gut inflammation [81, 82], which was consistent with our findings. Additionally, Romboutsia has been implicated in short-chain fatty acid production, which reduces systemic inflammation and macrophage infiltration [83]. Our results also showed a negative correlation between Romboutsia and serum inflammatory cytokine levels. In contrast, research on Dubosiella remains contradictory. While some studies suggest that Dubosiella mitigates Alzheimer's progression by producing palmitoleic acid, which has anti-inflammatory and anti-metabolic disorder properties [84], other studies report increased Dubosiella abundance in pesticide-induced Alzheimer's mouse models, accompanied by neuroinflammation [85]. Our results revealed a positive correlation between elevated Dubosiella abundance and hippocampal inflammation levels, suggesting a pro-inflammatory role in this context.

Given the diversity and complexity of the gut microbiota, we performed functional predictions of microbial communities and found that the LPS synthesis pathway was enriched in the active-phase fasting group. Quantitative analysis further confirmed that LPS levels in the cecum of the active-phase fasting group were statistically significantly elevated. LPS, a component of Gram-negative bacterial cell walls (e.g., Escherichia coli and Shigella), triggers immune responses when its levels exceed normal thresholds, particularly when gut integrity is compromised. LPS is frequently used to induce neuroinflammation models [86]. We observed a higher abundance of Escherichia-Shigella in the fasting group compared to the ND group (48.0 ± 5.7% vs. 18.0 ± 22.9%, p = 0.054), which likely contributed to the elevated LPS levels in the fasting group. LPS is one of the most potent activators of TLR 4. When LPS leaks into the serum, it binds to LPS-binding protein, which facilitates its transfer to CD14 and subsequently to the TLR 4/MD-2 complex [87]. This complex activates the MyD88-dependent signaling pathway, inducing the expression of pro-inflammatory mediators [88], which is one of the most canonical mechanisms underlying LPS-induced inflammation. Increased TLR 4 gene expression in the gut and hippocampus of the active-phase fasting group further supported the role of LPS in promoting inflammation.

Timing and composition of food intake are critical Zeitgebers for peripheral circadian clocks, and gut clock genes play an essential role in maintaining intestinal homeostasis. A mice study has shown that per1/2 double mutant mice, which have defective circadian rhythms, exhibit a loss of intestinal secretory cells, reduced mucin levels, weakened gut barrier protection, and exacerbated intestinal inflammation [89]. Similarly, the expression levels of circadian clock genes, including clock, cry1, cry2, per1, and per2, are reduced in the intestinal mucosa of patients with inflammatory bowel disease [90]. In contrast, the nampt gene is upregulated during acute intestinal inflammation [91]. Rorα has been shown to regulate T-cell infiltration in the colon and inhibit T-cell apoptosis, thereby prolonging the onset of chronic intestinal inflammation driven by pathogenic T cells [92]. Our findings also revealed that intestinal inflammation in IF dams was accompanied by changes in gut circadian gene expression, consistent with a previous study [93], and a correlation between the two was observed, suggesting a potential interaction. Although our results did not directly elucidate the mechanisms by which gut clock genes influenced intestinal inflammation, we identified a strong correlation between the two, suggesting a potential interplay that warrants further investigation.

Maternal pregnancy status is closely related to the growth and development of offspring. For example, maternal inappropriate eating-fasting cycles can lead to adverse pregnancy outcomes, including a reduction in offspring birth weight [60]. Increased maternal serum inflammation levels can impair early embryonic neurodevelopment, and their offspring may experience neurodevelopmental disorders and mental illnesses in childhood [94, 95]. Current research suggests that the maternal gut microbiota during pregnancy plays a crucial role in early embryonic development. Fetuses receive metabolic products from the mother through the placenta, which can be traced in the placenta and fetal tissues after maternal injection of LPS during pregnancy and exert effects via TLR 4 in the tissues [96]. High LPS levels during pregnancy lead to elevated inflammation in the offspring, impairing synaptic plasticity and causing autistic-like behaviors [97], which also depend on TLR 4 signaling. In our findings, higher levels of LPS were found in the placenta and brain of fetuses subjected to fasting during the active phase, and the mRNA levels of LPS receptor TLR 4 and several inflammatory cytokines were elevated, and impaired placental barrier function, elevated inflammatory cytokines, and reduced antioxidant capacity in the placenta [98], indicating that the fetuses received LPS from the dams and activated inflammatory pathways in tissues, which was strongly associated with the maternal microbiota, such as a negative correlation between Akkermansia and inflammatory cytokines in the fetal brain and a positive correlation with antioxidant capacity. Studies have shown that exogenous administration of Akkermansia can reduce systolic blood pressure in preeclampsia mice, promote fetal growth, and improve pathological changes in the placenta [99, 100]. Moreover, exogenous injection of Akkermansia into pregnant mice increases the number of active stem cells in the offspring's brain and gut, leading to faster recovery after disease [101]. Therefore, we believe that the maternal gut microbiota influences the fetuses during fasting in the active phase. This dietary habit reduces the abundance of beneficial bacteria such as Akkermansia, weakening its protective effects while increasing the abundance of harmful bacteria like Dubosiella, which in turn elevates the levels of Gram-negative bacteria producing LPS. Moreover, gut rhythmic genes, which are also involved in gut homeostasis, are disrupted by feeding times. Together, these factors activate the immune system, increasing the secretion of pro-inflammatory cytokines and raising the risk of disease in both the dam and her offspring.

Although the findings of this study are based on a rat model, increasing clinical evidence suggests that similar mechanisms may also exist in humans. Epidemiological surveys have shown that the prevalence of night eating during pregnancy is rising [32], and irregular maternal eating patterns are significantly associated with adverse pregnancy outcomes and emotional health issues, including preterm birth and fetal growth restriction [102]. More importantly, the colonization of gut microbiota involved in human immune and inflammatory processes originates in utero and continues throughout life [103]. Studies on human pregnancy have found that circadian rhythm disruption signals can be transmitted from the mother to the fetus [104, 105], and the gut microbiota also shows rhythmic fluctuations in response to feeding patterns [106]. Therefore, we propose that a shift in the maternal feeding window during pregnancy may affect early fetal development.

We suggest that pregnant women maintain a regular and reasonable eating schedule during gestation to prevent gut microbiota dysbiosis caused by poor dietary habits, which may negatively impact maternal and fetal health.

Limitations of the study

Our study has certain limitations due to its reliance on the rat model. There are notable differences between rats and humans in terms of feeding, sleep, and pregnancy cycles, which may lead to an overestimation of the findings. Additionally, the changes in the gut microbiota induced by dietary habits are based on the inherent microbiome of rats, which makes it challenging to extrapolate the results to humans. The way maternal microbiota and intestinal rhythms regulate fetal development remains unclear in this study. Future research will further explore the long-term effects of active-phase fasting on offspring into adulthood and use model organisms to clarify the complex relationship between gut microbiota and gut rhythmic genes.

Conclusions

In conclusion, we revealed the adverse effects of active-phase fasting on both the dams and their fetuses in rats, which occur through gut microbiota dysbiosis. We further found that during pregnancy, maternal gut imbalance leads to increased circulating LPS levels, which exert pro-inflammatory effects via TLR 4. This dietary pattern also altered the dynamics of intestinal circadian genes, which may also be associated with increased maternal intestinal inflammatory cytokines. Additionally, the elevated circulating LPS crossed the damaged placental barrier and entered the fetuses, leading to increased inflammation in the fetal brain. Given that human pregnancy involves a longer gestational period and more complex physiological activities compared to pregnant rats, we recommend that pregnant women maintain a proper eating schedule during pregnancy to avoid gut dysbiosis caused by dietary habits, which could negatively impact both maternal and fetal development.